PDF Archive

Easily share your PDF documents with your contacts, on the Web and Social Networks.

Share a file Manage my documents Convert Recover PDF Search Help Contact



Unit Commitment Towards Decarbonized Network Facing Fixed and Stochastic Resources Applying Water Cycle Optimization .pdf



Original filename: Unit Commitment Towards Decarbonized Network Facing Fixed and Stochastic Resources Applying Water Cycle Optimization.pdf
Title: Unit Commitment Towards Decarbonized Network Facing Fixed and Stochastic Resources Applying Water Cycle Optimization
Author: Heba-Allah I. ElAzab, R. A. Swief, Noha H. El-Amary and H. K. Temraz

This PDF 1.5 document has been generated by LaTeX with hyperref package / pdfTeX-1.40.18, and has been sent on pdf-archive.com on 16/08/2018 at 11:59, from IP address 193.137.x.x. The current document download page has been viewed 70 times.
File size: 3.6 MB (21 pages).
Privacy: public file




Download original PDF file









Document preview


energies
Article

Unit Commitment Towards Decarbonized Network
Facing Fixed and Stochastic Resources Applying
Water Cycle Optimization
Heba-Allah I. ElAzab 1
1
2
3

*

ID

, R. A. Swief 2 , Noha H. El-Amary 3, *

ID

and H. K. Temraz 2

Faculty of Engineering, Ahram Canadian University(ACU), Giza 12573, Egypt; hebaelazab2013@gmail.com
Faculty of Engineering, Ain Shams University, Cairo 11517, Egypt; rania.swief@gmail.com (R.A.S.);
htemraz@gmail.com (H.K.T.)
Arab Academy for Science, Technology and Maritime Transport (AASTMT), Cairo 2033, Egypt
Correspondence: noha_helamary@ieee.org or noha_helamary@hotmail.com; Tel.: +20-100-471-8562

Received: 21 March 2018; Accepted: 29 April 2018; Published: 3 May 2018




Abstract: This paper presents a trustworthy unit commitment study to schedule both Renewable
Energy Resources (RERs) with conventional power plants to potentially decarbonize the electrical
network. The study has employed a system with three IEEE thermal (coal-fired) power plants
as dispatchable distributed generators, one wind plant, one solar plant as stochastic distributed
generators, and Plug-in Electric Vehicles (PEVs) which can work either loads or generators based
on their charging schedule. This paper investigates the unit commitment scheduling objective
to minimize the Combined Economic Emission Dispatch (CEED). To reduce combined emission
costs, integrating more renewable energy resources (RER) and PEVs, there is an essential need
to decarbonize the existing system. Decarbonizing the system means reducing the percentage of
CO2 emissions. The uncertain behavior of wind and solar energies causes imbalance penalty costs.
PEVs are proposed to overcome the intermittent nature of wind and solar energies. It is important
to optimally integrate and schedule stochastic resources including the wind and solar energies, and
PEVs charge and discharge processes with dispatched resources; the three IEEE thermal (coal-fired)
power plants. The Water Cycle Optimization Algorithm (WCOA) is an efficient and intelligent
meta-heuristic technique employed to solve the economically emission dispatch problem for both
scheduling dispatchable and stochastic resources. The goal of this study is to obtain the solution for
unit commitment to minimize the combined cost function including CO2 emission costs applying
the Water Cycle Optimization Algorithm (WCOA). To validate the WCOA technique, the results
are compared with the results obtained from applying the Dynamic Programming (DP) algorithm,
which is considered as a conventional numerical technique, and with the Genetic Algorithm (GA) as
a meta-heuristic technique.
Keywords: plug-in electric vehicles (PEVs); water cycle optimization algorithm (WCOA); quadratic
programming; combined economic emission dispatch (CEED)

1. Introduction
The unit commitment study integrating stochastic and disputable resources is a rich topic with
different aspects and branches, but all those branches have their scope in the main theme of the work.
The guidelines of the introduction are divided into the following points:




Unit commitment importance and aim of the study;
The reasons for selecting the objective function governing the unit commitment study, emission
cost reduction;

Energies 2018, 11, 1140; doi:10.3390/en11051140

www.mdpi.com/journal/energies

Energies 2018, 11, 1140






2 of 21

The advantages and disadvantages of integrating RERs into the study goals;
The integration of PEVs and their advantages and disadvantages for achieving the quality of the goals
A state of art in the unit commitment area and the optimization technique applied;
The contribution and the structure of the paper

Unit commitment is a vital study required to ensure the hourly energy supply requirements.
The unit commitment focuses on minimizing the production cost, which mainly depends on the fuel
cost value. However, with the increase of fuel cost, the CO2 emissions will increase. The goal of the
study is to decarbonize the CO2 limit in electrical power system networks, which means reducing the
amount of CO2 emissions. To reduce CO2 emissions while supplying the required demands, integrating
more Renewable Energy Resources (RER) will cause a conflict problem as a result of increasing the
amount of CO2 emissions, which causes the earth temperature to rise. The unit commitment problem
is a complicated optimization problem, from the objective function point of view or its constraints [1–5].
The unit commitment problem is defined by scheduling the generation power attained from various
power resources. Conventional and intelligent programming techniques are used to solve the unit
commitment problem by achieving priority list combination of the generating units, so that the
combined emission production cost can be minimized. Many conventional techniques have been
applied to solve the unit commitment problem such as the mixed integer optimization [3] and Lagrange
method [2,4]. One of the most effective and robust conventional methods is dynamic programming
(DP), which is based on the available combinations of resources. This method proves that it is simple
and fast and provides autocratic and effective solutions [5]. Due to the large number of resources
integrated into the electric grid and the related number of constraints, the need for fast computational
technique is urgent. The main purpose of this paper is to provide an optimization framework by
scheduling the wind and solar energies and PEVs (load-generator) as stochastic distributed generators
and dispatchable distributed generators. This coordination can handle the imbalances of intermittent
Renewable Energy Resources (RERs) and encourage PEVs passengers to take part in the demand
response while optimal hourly prices are determined.
On the other hand, the international communities seek to prevent temperature rise more than
2 degrees Celsius by generating more energy from domestic resources which can be cost-effective
and replaced or renewed without contributing to climatic change or having environmental impacts.
Burning fossil fuels such as coal, natural gas and oil, which exhaust ash and gaseous pollutants such
as carbon oxides (CO and CO2 ) nitrogen oxides (NOx )) and sulphur dioxide (SO2 ) . . . etc. Electricity
production is roughly responsible for half of the greenhouse gas emissions (GHGEs). In fact, it is
expected that the fossil-fuel power plants planned to be built, will emit tens of billion tons of carbon
dioxide over their expected lifetime, compared with the annual emissions of all fossil-fuel power plants
that were operating in the preceding years. Excluding these fossil fuel power plants early is achievable,
but the cost comparison for decision-makers who compare fossil fuels with clean energy resources
remains a critical issue. Long-term planning will lead to stabilizing climatic changes and achieving
zero emissions. Therefore, the goal is decreasing the emissions in the upcoming decades to attain zero
net emissions by the end of this century, which can be achieved by applying the unit commitment study.
Carbon–neutral electricity can be produced by using renewable resources (windmills, photovoltaic
power, concentrated solar power, nuclear power, large dams and small hydropower) and fuel shifting
technologies such as electric and plug-in hybrid vehicles in the transportation sector [6].
Renewable energy resources (RERs) in such systems integrate with conventional power plants
seeking to achieve potential decarbonization of the electrical system. RERs consist of low-carbon base
load generation technologies such as nuclear and fossil fuels with carbon capture and sequestration,
along with more modest contributions (25%) from wind (whether on-shore or off-shore), and solar
(whether photovoltaic or solar thermal cells) [7]. Solar integration can help in improving and reducing
the pollution limits obtained from fossil fuel substations. Encouraging residential customers to use
PV solar microgeneration can save 3.5% of energy consumption in addition to reducing the overall
cost by 75% lower than the models without PV [8–11]. Energy storage devices (ESDs) with PV panels

Energies 2018, 11, 1140

3 of 21

can contribute to saving 12% of the energy cost. Hence, this encourages end users with Home Energy
Management Systems (HEMSs) to reduce or shift their electricity consumption patterns in response to
a price or grid condition signal [12]. By contributing to customers’ payment for the cost of generation
as well as for the transmission, distribution and indirect cost of environmental cleanup and health
effects, these renewable resources may become attractive when customers face the actual price of
electricity production [13].
Because of the uncontrollable behavior of wind speed and solar insolation, the output of wind
and solar power generation is unpredictable. Swings in generation of wind and solar energy between
oversupply and undersupply will lead to estimations of the hourly contributions from wind and PV as
being insufficient to satisfy renewable portfolio standards and overwhelm any conceivable storage
strategy. Consequently, wind and solar energies will compensate a part of the load demand. It is
necessary to handle the uncertain behavior of these renewable resources which may cause operational
risk of power system [14]. Controlling both active and reactive power independently is done to overcome
the stochastic nature of wind energy profile by using voltage source inverters (VSIs) based on FACTS
devices (STATCOM) + battery energy storage systems (BESS) which will promote the transient and
dynamic stability and minimize sub-synchronous oscillations [15]. Battery storage technologies (i.e.,
pumped hydroelectric storage (PSH)) may have a vital role in solving imbalances due to seasonal
swings in the generation output of wind and solar power plants. Furthermore, battery storage will
decrease energy costs and support dependence on renewable energy for the off-grid areas (remote
locations) not connected to the grid and therefore resort to using expensive imported fuels [14,16].
Exchanging the old generation with the new RERs technologies has proved its efficiency in
many cases. For example, in Tasmania, the diesel consumption of King Island has fallen to 50% after
exchanging renewable energy and battery storage for the conventional power plants [17]. Another
example is found in an existing network of Masirah Island, where the hybrid energy system is
composed of fuel-fired, photovoltaic and wind generator units is considered the most economically
feasible combination. Despite the high capital costs of this combination, it provides the cheapest
operating cost, energy cost and hence the lowest net running cost. This combination of units also
enhances the voltage profile of the system [18]. In 2013, as leading global steps in renewable energy,
the German government encouraged utility and smaller-scale battery storage by introducing many
incentives for households, companies, local authorities and community organizations. Households
and businesses could access grants for 30% of the upfront installation costs to install new solar PV and
storage systems [14,16]. The energy storage system (ESS) can provide an alternative by satisfying peak
demand to achieve load levelling and peak reduction of up to 8% by using GA in the UK distribution
network with a peak capacity violation [19].
Plug-in electric vehicles are expected to become widely common in the upcoming decades.
2010 witnessed release of the first plug-in hybrid Chevy Volt made by General Motors and the
all-electric Leaf was released by Nissan. In the first half of 2017, the sales of new vehicles were
50 percent higher than sales in 2016 and it is predicted that new global PEV sales will exceed this
percentage compared with traditional fuel vehicle sales by 2050 [14,16]. Encouraging passengers to
use PEV to shift peak load will help in supplying load in uncertainty situations [20]. Reasons for the
optimistic forestation of plug-in electric vehicles include:






They are quiet due to reducing the tailpipe emissions and air pollutants produced by gasoline or
diesel-powered engines which harm the heart and lung health for the people living near roads;
They require less maintenance;
Recharging is cheaper than refueling with petrol which is a depleted energy resource, leading to
less reliance on fossil fuels if powered by renewable energy;
The vehicle battery can also be used for household electricity storage [21,22].

Plug-in electric vehicles (PEVs) can be considered as a battery storage based on the concept of
battery energy storage systems in which vehicles operate with an electric motor that can be powered by

Energies 2018, 11, 1140

4 of 21

an external electrical source. From a grid perspective, two basic types of EVs are typically indicated to
BEVs as the battery is considered a primary source of power in EVs, or plug-in hybrid electric vehicles
(PHEVs) are combined with a secondary drive option. Some models feature a gasoline-powered engine
such as the internal combustion engine (ICE). The battery capacity ranges from less than 10 kWh to over
80 kWh. However, the battery charging and discharging process in PEVs may cause sharp, unexpected
spikes in electric power consumption and lead to potential grid issues [23]. PEVs support grid
operation by providing distributed energy storage in the form of vehicle to grid (V2G). After optimally
determining the appropriate size for renewable resources and storage devices [24,25], the optimal
scheduling takes place. By the optimal intelligent scheduling of PEVs, V2G potentially provides grid
generation to reduce the intermittency and uncertainty of renewable resources such as wind and solar
power [26].
Charging of PEVs can be scheduled at night as well as during weekends, when electricity prices
are comparatively low and when vehicles are not used. However, some PEVs charging will be needed
during the daylight and even during peak demand or rush hour intervals when the grid already
provides the maximum electric power capacity [27]. There are two methods of the charging strategy of
the scheduling PEVs, namely, the reactive strategy and the proactive strategy. The reactive strategy:
As soon as the PEVs have been plugged-in, some unnecessary loads like heaters and dryers are
postponed for decreasing the base load. Non-critical but high consuming loads can be turned off by
the center controller until the battery of PEVs is being fully charged. However, the domestic loads
cannot be easily controlled. The proactive strategy estimates the averages charging scenario and the
future capacity to avoid overloading with the day-ahead load profile. During fast charging process,
the proactive method may cause a sudden spike, but may take a shorter charging interval. The reactive
method can handle the deviation risk, but charging process occur in more time [28].
Due to the complexity of the problem, there is a crucial need for a powerful optimization technique
to find the optimal solution which satisfies the objective and the constraints [29–33]. The water
cycle optimization algorithm is applied to solve the economical emission dispatch unit commitment
problem. The algorithm is considered a new meta-heuristic technique where the obtained results will
be compared with another heuristic technique such as the genetic algorithm (GA) and traditional
technique as dynamic programming (DP). The contribution of this paper mainly appears in scheduling
the PEVs as load-generators, the construction of the optimization function including the emission
cost not as a penalty [2], but as a cost that should be paid, and applying the WCOA in a new field to
validate its performance with respect to other conventional and meta-heuristic techniques.
The paper is divided into seven sections. Section 2 presents an overview of the unit commitment
formulation and constraints and outlines the steps and procedures of the Dynamic Programming
technique. Section 3 delivers the main rules and concepts about the Water Cycle Algorithm. Section 4
represents the data of the system under study. Simulation results are illustrated in Section 5. Section 6
summarizes the discussion main points, while Section 7 presents the paper’s conclusions. Also,
a summary of all variables and acronomies is displayed.
2. Unit Commitment Formulation
Unit commitment formulation in this study can be considered as a multi-objective, single function
representation optimization problem which aims to minimize both the operating cost and the emission
cost. Not only the renewable energy resources (RER) such as wind and solar energies will be added to
the electric industry to decrease the emission, but the PEV as well. The PEV will be used to reduce
both the cost and the emission in the electricity and transportation sectors. The operating cost of the
thermal units includes the fuel cost depending on the amount of fuel consumption by the thermal
generating units and the start-up cost depending on the temperature of the boilers:
NG

Fuel thermal cos t =

∑ Ai + Bi PG i + Ci P2G i

i=1

(1)

Energies 2018, 11, 1140

5 of 21

where; PG i is the output power of each thermal unit “i” at each hour. A, B, C are the coefficients of
a quadratic fuel cost function of each thermal generating unit. NG is the number of conventional
thermal units.
A linear model is used to evaluate the fuel cost of wind and solar energy:
Fuel cos t (wind/solar) = (wind/solar) price × Pwind/solar

(2)

where; Pwind/solar is the output power from wind or solar plants at each hour. (wind/solar) price is
the coefficient of a linearized fuel cost function of wind or solar plants at each hour:
NG

Emission cos t =

∑ α × 103 × PG i × β

(3)

i =1

where α is CO2 emission factor that represents the ratio between the quantity of gas emitted (in ton)
per unit of energy production (in kWh). β is the emission penalty factor in voluntary markets for
planning purposes which is around 10–15 $/ton CO2 by the end of 2017 [27]. β is defined to be the
average of carbon prices, according to the World Bank’s annual Carbon Pricing Watch Report 2017.
The CO2 emission factor (α) is shown in Table 1, together with the emission factor of energy
resources for both burnt fossil fuels (natural gas, fuel oil and coal) and renewable energy resources
(wind, hydropower and solar photovoltaic).
Table 1. CO2 emission factor “α” for different energy resources [34].
Energy Resource

CO2 Emission Factor (Ton/kWh)

Wind
Hydro
Solar
Natural Gas
Fuel oil
Coal

21.0 × 10−6
15.0 × 10−6
6.00 × 10−6
5.99 × 10−4
8.93 × 10−4
9.55 × 10−4

PEVs Operating Constraints
Conventional thermal (coal-fired) units, RERs and PEVs which smartly operate as loads, energy
storages or small portable power plants (energy sources) should meet and supply the whole load
demand (hour) and the system losses which are described as follows:



If PEVs are operated as an energy resource or a small power plant:
N

∑ PG i (hour)

i =1

+Pwind (hour) + Psolar (hour) +

NV2G (hour)



j =1

h
i
η PPEVj (hour) ΨPres (hour) − Ψdep (hour)

(4)

= Demand(hour) + Reserve (hour)



If PEVs are added to the demand as loads:
N

∑ PG i (hour)

i =1

+Pwind (hour) + Psolar (hour) = Demand(hour) + Reserve (hour)
+

NV2G (hour)



j =1

h
i
η PPEVj (hour) ΨPres (hour) − Ψdep (hour)

(5)

where; PG i is the output power of each thermal unit “i” at each hour. PPEVj is the power of each
vehicle j, η is plug in vehicle system efficiency. NV2G is number of vehicles that are connected to
the network at this hour. N is number of units that are on in the unit commitment problem at
each hour. ΨPres & Ψdep are the present and the departure state of charge (SOC) respectively.

Energies 2018, 11, 1140

6 of 21

Some Other Constraints
The generating power limit for each generating unit should be within the minimum and
maximum limits:
PG imin ≤ PG i ≤ PG imax
(6)
where; PG imin & PG imax are the minimum and maximum output limits of the i-th thermal unit
respectively, considering ramp up/down rate, minimum up/down time and the spinning reserve of
the system at each hour, which will be required to preserve the reliability and the adequacy of the
system. Only the recorded vehicles will contribute to the smart operation according to predetermined
scheduling intervals:
24



NV2G (hour) = NV2G max

(7)

hour=1

where; NV2G is the number of vehicles connected to the network at this hour. NV2G max is the total
number of vehicles in the network.
To avoid loss of battery life:
Ψmin (hour)PPEVj ≤ PPEVj (hour) ≤ Ψmax (hour) PPEVj

(8)

where; Ψmin is the depletion of storage energy at minimum level. Ψmax is the charging up to maximum
level.
The multi-objective single-representation objective function for Cost Emission Economical
Dispatch (CEED) optimization that is required to be minimized in the smart grid is expressed as follows:
CEED = Minimize total cos t = min{fuel cos t, start − up cos t, emission}
N

= ∑

24

∑ {Fuel thermal cos t(PG i (hour)) + Fuel cos t (wind/solar) (hour)+

(9)

i =1 hour =1

start − up cos ti(hour) × (1 − Ui (hour − 1)) + Emission cos t (Pi (hour))} × Ui (hour)
where, Ui (hour) is on/off state of each unit i.
2.1. Unit Commitment Solution Using Dynamic Programming Technique
In 1957, Bellman interpreted the theory of dynamic programming (DP). Combination of the
generating units is considered as the states which need to be determined. Searching for the optimum
solution is achievable for each time interval (hour) in a forward or backward direction.
In forward direction of dynamic programming, the optimistic economic combination of the
minimum accumulated cost starts at the last stage and ends at the initial stage. The stages form
the intervals of the study problem considered as T = 24 h. The combination of each generating unit
forwards one hour, then the arrangement is scheduled and stored for each hour. Eventually, at the final
hour, the best path of the most economical schedule of the generating power for each unit is attained
by backward pedaling. The main advantage of this method is that the dimensionality of the problem
can be significantly ignored, so that dynamic programming can get the best path of the minimum cost
for running NG units [35,36].
2.2. Step by Step Tracking for Dynamic Programming
The dynamic programming approach is based on the probabilities of generators to supply the
load at a certain hour, satisfying the constraints to achieve the objective function which minimizes the
combined emission costs. The dynamic programming algorithm for unit commitment problem, in its
elemental form, tackles every probable state in every interval. The algorithm is based on the posterior
repeating equation:
Total_ cos t(t, I) = min [prod_ cos t(t, I) + s_ cos t(t − 1, L : t, I) + total_ cos t(t − 1, L)]

(10)

the combined emission costs. The dynamic programming algorithm for unit commitment problem,
in its elemental form, tackles every probable state in every interval. The algorithm is based on the
posterior repeating equation:
Energies 2018, 11, 1140

Total_cost(t, I) = min [ prod_cost(t, I) + s_cost(t − 1, L: t, I) + total_cost(t − 1, L)]

7 of 21

(10)

where total_cost(t,I) is the least total cost required to arrive at state (t,I); prod_cost(t,I) is the
where total_cost(t,I)
is the(t,I);
leastand
total
cost required
to arrive
state (t,I);cost
prod_cost(t,I)
production
production
cost at state
s_cost(t
− 1,L: t,I)
is the at
transition
from state is
(t the
− 1,L)
to state
cost
at
state
(t,I);
and
s_cost(t

1,L:
t,I)
is
the
transition
cost
from
state
(t

1,L)
to
state
(t,I).
(t,I).
Strategyin
inforward
forwarddynamic
dynamicprogramming
programmingisisdefined
definedasas
the
transition
path
from
state
Strategy
the
transition
oror
path
from
oneone
state
at
at
a
given
hour
to
a
state
at
the
next
hour.
The
state
(t,
1)
is
the
I-th
combination
in
hour
t,
where
N
a given hour to a state at the next hour. The state (t, 1) is the I-th combination in hour t, where N is
is
the
number
of
strategies
to
save
at
each
step
and
X
is
the
number
of
states
to
search
each
period.
the number of strategies to save at each step and X is the number of states to search each period. The
The maximum
or is
N 2isN 2−N 1.
− The
1. The
presumption
stepbybystep
stepprocedure
procedureand
and the
the
maximum
valuevalue
of XoforX N
presumption
forfor
thethe
step
flowchart
of
the
dynamic
programming
method
are
expressed
in
Figure
1
and
explained
as
follows:
flowchart of the dynamic programming method are expressed in Figure 1 and explained as follows:
(a) AAstate
stateisiscomposed
composedof
ofan
anarrangement
arrangementof
ofgenerating
generatingunits
unitswith
withonly
onlyaccurate
accurateunits
unitsin
inservice,
service,
(a)
operatingat
ataatime
timewhile
whilethe
theremaining
remainingunits
unitsare
areoff-line.
off-line.
operating
(b) The
The start-up cost
unit
is fixed
and independent
of the time,
or not theor
generating
(b)
costofofeach
each
unit
is fixed
and independent
of thewhether
time, whether
not the
unit is in off
state.
generating
unit
is in off state.
(c)
(c) No
Nocost
costisisinvolved
involvedfor
forthe
theshutdown
shutdownof
ofthe
theunit.
unit.
(d)
interval,the
theorder
order
priority
is firm
a amount
small amount
of should
powerbe
should
be in
(d) At
At each interval,
of of
priority
is firm
and aand
small
of power
in operation.
operation.
START

Total_cost(t,I) = min [ prod_cost(t,I) + s_cost(t-1,L:t,I) ] [ L ]

t=t+1

L = “N” feasible states in interval t - 1

Total_cost(t,I) = min [ prod_cost(t,I) + s_cost(t-1,L:t,I) + total_cost(t-1,L)]

Save in lowest cost strategies

NO
t = T , last hour?
YES
Trace the optimal schedule

END

Figure
Figure1.
1.AAflowchart
flowchartfor
forthe
theforward
forward dynamic
dynamic programming
programming algorithm.
algorithm.

By increasing number of resources in the network, the number of permutations and
By increasing number of resources in the network, the number of permutations and combinations
combinations will become hectic. So the crucial need to find the optimal solution in an intelligent
will become hectic. So the crucial need to find the optimal solution in an intelligent way is important
way is important to save time and complexity. The water cycle algorithm, genetic algorithm
to save time and complexity. The water cycle algorithm, genetic algorithm compared to the quad
compared to the quad programming is exposed showing the power and the efficiency of the applied
programming is exposed showing the power and the efficiency of the applied algorithm.
algorithm.
3. The Water Cycle Optimization Algorithm (WCOA)
In 2013, the water cycle algorithm (WCA) is considered as a meta-heuristic technique. The new
technique is used to obtain the optimal solutions for problems. The main concept of constructing the

Energies 2018, 11, x FOR PEER REVIEW

8 of 22

3. The Water Cycle Optimization Algorithm (WCOA)
Energies 2018, 11, 1140

8 of 21

In 2013, the water cycle algorithm (WCA) is considered as a meta-heuristic technique. The new
technique is used to obtain the optimal solutions for problems. The main concept of constructing the
water
cycle
algorithm
is inspired
by the
natural
phenomena
of the
water
cycle.
flow
of the
rivers
water
cycle
algorithm
is inspired
by the
natural
phenomena
of the
water
cycle.
TheThe
flow
of the
rivers
and
streams
into
the
sea
occurs
as
in
real
life
as
illustrated
in
Figure
2.
Water
is
formed
from
and streams into the sea occurs as in real life as illustrated in Figure 2. Water is formed from rain,rain,
other
streams
high
themountains,
mountains,when
whensnow
snowand
and glaciers
glaciers melt. Water
other
streams
or or
high
upup
ininthe
Water travels
travelsdownhill
downhilland
andforms
formsaariver
riverororaastream.
stream.Rivers
Riversand
andstreams
streamsflow
flowdownhill
downhillon
ontheir
theirjourney
journeytowards
towardsthe
thesea.
sea.

Figure
2. A2.simplified
diagram
of the
cycle
(the(the
hydrologic
cycle)
[37].[37].
Figure
A simplified
diagram
of water
the water
cycle
hydrologic
cycle)

3.1.3.1.
Hydrologic
Cycle
Hydrologic
Cycle
TheThe
hydrologic
cycle
consists
of: of:
hydrologic
cycle
consists




Evaporation process: Because of high temperatures, water in lakes and rivers evaporate.

Evaporation process: Because of high temperatures, water in lakes and rivers evaporate.
Transpiration process: During photosynthesis, plants transpire and give off water.

Transpiration process: During photosynthesis, plants transpire and give off water.
Condensation process: The water evaporated from rivers and the water transpired by trees
Condensation process: The water evaporated from rivers and the water transpired by trees

generate clouds when such water is condensed into the colder atmosphere.
generate clouds when such water is condensed into the colder atmosphere.

Precipitation process: The water is released back to the earth in the form of rain

Precipitation process: The water is released back to the earth in the form of rain

Percolation process: When rain falls and glaciers melt, the water is reserved beneath the
Percolation
process: Whenaquifer,
rain falls
and glaciers
melt,inthe
water
beneath
• ground.
The groundwater,
flows
downward
the
sameis reserved
way water
flowsthe
onground.
the
The
groundwater,
aquifer,
flows
downward
in
the
same
way
water
flows
on
the
ground
surface
ground surface completing the hydrologic cycle [37].
completing the hydrologic cycle [37].
3.2. The Proposed WCOA
3.2. The Proposed WCOA
Similar to other meta-heuristic algorithms, rain or precipitation forms raindrops which are
Similar to other meta-heuristic algorithms, rain or precipitation forms raindrops which are
considered as the initial population in the proposed algorithm. When the best raindrop is found, the
considered as the initial population in the proposed algorithm. When the best raindrop is found,
best individual (raindrop) is selected as a sea. Many good raindrops are selected as a river and the
the best individual (raindrop) is selected as a sea. Many good raindrops are selected as a river and the
rest are considered as streams which flow into the rivers and the seas. Using the population based
rest are considered as streams which flow into the rivers and the seas. Using the population based
meta-heuristic techniques “Raindrop” is a single solution in an array of 1 × Nvar, where Nvar is a
meta-heuristic techniques “Raindrop” is a single solution in an array of 1 × Nvar , where Nvar is a
dimensional optimization problem or number of the design variable [38,39]:
dimensional optimization problem or number of the design variable [38,39]:
Raindrop = [ X1 , X2 , X3 … XN ]
(11)
Raindrop = [X1 , X2 , X3 . . . XN ]
(11)
A population of raindrops is generated as a matrix of raindrops of size Npop × Nvar, where
Npop is A
the
number ofofpopulation
asgenerated
per the following
Equation
(12): of size N
population
raindrops is
as a matrix
of raindrops
× N , where N
pop

is the number of population as perRaindrop
the following
Equation
(12):
1
X11
X12
Raindrop
2

2
X1
X22
Population of raindrops =Raindrop1 ⋮
 = 1


X1 ⋮
X12 ⋮

Raindrop2 ⋮ 
Npop
Npop

 
X
X22
2



X[1 1
X
..
[Raindrop
Npop

 ]
2
 =
Population of raindrops = 
.

 
..
..

 
.
.
..

 


.
N
N




RaindropNpop

X1

pop

X2

pop

var

X13 … X1Nvar
2
X32 … XNvar
⋮ X1 ⋮ . . . ⋮X1
Nvar
Npop
Npop 3
X3 X2 … . . .XNvar
X2 ]
3

..
.
Npop

X3

..
.
...

Nvar

pop

(12)

..
.
Npop

XNvar









(12)

Energies 2018, 11, 1140

9 of 21

where (X1 , X2 , X3 , . . . , XNar ) are the decision variable values which can be defined as floating
point number (real values) for continuous and discrete problems. The cost function of raindrops
is represented in the following Equation (13):


Ci = Costi = f X1i , X2i , X3i , . . . , XiN , i = 1, 2, 3, . . . , Npop .
(13)
Seas and rivers are chosen as minimum values (the best individuals). Nsr is defined as the
summation of the number of rivers, which is considered as a user parameter and a single sea. The other
raindrops (population) flow either to the rivers or directly to the sea as per the following equations:
Nsr = Number of Rivers + 1, where 1 is for one sea

(14)

NRaindrops = Npop − Nsr

(15)

The intensity of the flow determines how to assign raindrops to the rivers and the sea as follows:

(
)
Cost



n
NSn = round N
(16)
× NRaindrops , n = 1, 2, . . . , Nsr .
∑ sr Costi
i =1
where NSn is defined as the number of streams, which travels towards certain rivers or the sea. Figure 3
describes the WCA optimization process in which X is the distance between the stream and the river,
can be randomly chosen as follows:
X ∈ (0, C ≤ d), 1 < C < 2

(17)

where C is between 1 and 2.; and d is defined as the current distance between stream and river.
The value of X in Equation (17) is set according to a randomly distributed number whether (uniformly
or in an appropriate distribution) between 0 and (C × d). Enabling C > 1, streams are permitted to flow
in various directions towards rivers. This concept can explain rivers flowing into the sea. Therefore,
from the point of the exploitation phase in the WCOA, the new position for streams and rivers can be
obtained as follows:


i +1
i
i
i
Xstream
= Xstream
+ rand × C × XRiver
− Xstream
(18)


i +1
i
i
i
XRiver
= XRiver
+ rand × C × XSea
− XRiver
(19)
where rand is a randomly distributed number in a uniform way between 0 and 1. If the solution
obtained by a stream is better than its linking river, the positions of river and stream are swapped (i.e.,
stream becomes river and river becomes stream). The same exchange can occur for rivers and the sea
as shown in Figure 4. In the evaporating process, the assumption of evaporating water as streams or
rivers is to bypass enclosing in local optima. Therefore, the following Pseudo code is to determine
whether or not river flows into the sea.
If |X iSea − X iRiver| < dmax ; i = 1, 2, 3, . . . , (Nsr , the evaporation and raining process will
be ended.
Where dmax is considered as a small number and its value is near to zero. Therefore, the distance
between a river and sea should be less than dmax . It indicates that the river is linked to the sea, the
evaporation process occurs; and the precipitation process (raining) is applied after some adequate
evaporation. To get the optimum solution, dmax will control the search intensity close to the sea.
i +1
i
dmax
= dmax


dimax
max − iteration

(20)

i
where if dmax
is large, the search is decreased while a small value will be helpful to search close to the
sea. In the precipitation process, the new position of the new streams formed by the new raindrops is
expressed as follows:
Xnew
(21)
Stream = LB + rand × (UB − LB)

Energies 2018, 11, 1140

10 of 21

whereEnergies
LB and
UB are represented as lower and upper boundaries of the designed problem
in the
2018, 11, x FOR PEER REVIEW
10 of 22
Energies
2018, 11, x FOR
REVIEW
10 of for
22 the
proposed
algorithm
andPEER
from
the point of the exploration phase. Equation (22) is applied only
constrained
problems
the streams
directly
travel toward
the sea: problem in the
where LB
and UB to
areenhance
represented
as lower which
and upper
boundaries
of the designed

where LB and UB are represented as lower and upper boundaries of the designed problem in the
proposed algorithm and from the point of the exploration phase. Equation (22) is applied only for
proposed algorithm and fromnew
the point of the√
exploration phase. Equation (22) is applied only for
the constrained problems to X
enhance
theXstreams
directly
× randn
Nvar ) toward the sea:
(22)
(1, travel
Sea + µwhich
Stream =
the constrained problems to enhance
the streams
which directly travel toward the sea:
new
Xnew
(22)
Stream = XSea +√µ × randn (1, Nvar)
XStream
XSeathe
+õrange
× randn
(1, Nvar) area close to the sea and the
(22)value
where µ is considered as a coefficient
to=get
of searching

where
µ is considered
as a the
coefficient
to
get
the range
of searching
area
close to the sea
and the
value
should
be smaller
to improve
searchto
inget
a smaller
area.
From the
mathematical
point
where
µ is considered
as a coefficient
the range
of searching
area
close to the sea
andof
theview
value µ is
should
be
smaller
to improve
the search
inthe
a smaller
area. From
the mathematical
point
of view
õ
described
asbe
thesmaller
standard
deviation
µ isin
variance.
best
of µ is 0.1point
to get
optimum
should
to improve
theand
search
a smaller
area.The
From
thevalue
mathematical
of the
view
õ
is described as the standard deviation and µ is the variance. The best value of µ is 0.1 to get the
solution
(sea) [32].asFor
the algorithm,
solution
obtained
is described
theterminating
standard deviation
and µ isthe
thebest
variance.
Theisbest
value at
of the
µ ismaximum
0.1 to get number
the
optimum solution (sea) [32]. For terminating the algorithm, the best solution is obtained at the
optimum
solution
(sea)
[32].
For
terminating
the
algorithm,
the
best
solution
is
obtained
at
the
of iteration, “ε” (CPU time) is a small and non-negative number which is the allowable tolerance
maximum number of iteration, “ε” (CPU time) is a small and non-negative number which is the
maximum
number ofsolutions.
iteration, “ε” (CPU time) is a small and non-negative number which is the
between
two successive
allowable
tolerance between
two successive solutions.
allowable tolerance between two successive solutions.

Figure 3. Schematic view for water cycle algorithm [38].

Figure
3. 3.
Schematic
cyclealgorithm
algorithm
[38].
Figure
Schematicview
viewfor
for water
water cycle
[38].

Figure 4. The position of the stream (raindrop) and the river replacement [38]. The star is the river

Figure
4. position
The position
of the stream
(raindrop)
and
river
replacement[38].
[38].The
The star
star is
is the
the river
Figure
4. The
(raindrop)
and
thethe
river
replacement
river and
and
the black
circleofisthe
the stream
best among
other streams.
and
the
black
circle
is
the
best
among
other
streams.
the black circle is the best among other streams.
3.3. Step by Step Tracking and the Water Cycle Optimization Algorithm Flowchart
3.3. Step by Step Tracking and the Water Cycle Optimization Algorithm Flowchart
3.3. Step
Step1: Tracking
the Water Cycle
Optimization
Flowchart
 byStep
initialize and
the parameters
of WCOA:
Nsr, dmaxAlgorithm
, Npop, max_iteration.

Step 1: initialize the parameters of WCOA: Nsr, dmax, Npop, max_iteration.

Step
2: create a parameters
random generation
of initial
generate the initial raindrops,

Step
1:
initialize
of WCOA:
Nsr , population
dpopulation
,and
max_iteration.
max , Npop and

Step
2: createthe
a random generation
of initial
generate the initial raindrops,
rivers and sea by using Equations (12), (14) and (15).

Step rivers
2: create
random
generation
initial
and generate the initial raindrops, rivers
andasea
by using
Equationsof
(12),
(14) population
and (15).

Step 3: determine the cost of each stream (raindrops) by using Equation (13).
 sea
Stepby
3: using
determine
the cost(12),
of each
stream
(raindrops) by using Equation (13).
and
Equations
(14)
and
(15).

Step 4: evaluate the intensity of flow for rivers and sea by using Equation (16).

Step
4: evaluatethe
the intensity
of flow
for rivers and seaby
byusing
using Equation
Equation (16).

Step
3:
determine
stream

Step
5: evaluate thecost
flowofofeach
streams
into (raindrops)
the rivers by using
Equation (18). (13).

Step 5: evaluate the flow of streams into the rivers by using Equation (18).

Step
6: evaluate
the flow of
intorivers
the sea
which
hasusing
the most
downhill

Step
4:
evaluate
the intensity
of rivers
flow for
and
sea by
Equation
(16).place by using

Step 6: evaluate the flow of rivers into the sea which has the most downhill place by using
Equation
(19).

Step Equation
5:
evaluate
the flow of streams into the rivers by using Equation (18).
(19).

Step 7: replace the positions of river and stream which achieves the best solution as described in

Step
6: evaluate
of rivers
seawhich
which
has the
downhill
place byinusing

Step
7: replacethe
the flow
positions
of riverinto
and the
stream
achieves
themost
best solution
as described
Figure 4.
Equation
(19).
Figure
4.

Energies 2018, 11, 1140








11 of 21

Step 7: replace the positions of river and stream which achieves the best solution as described in
Figure 4.
Step
8:2018,
the 11,
same
in step
7, replace the position of river with the sea which achieves
the
Energies
x FORas
PEER
REVIEW
11 of
22 best
solution. The river may provide a better solution than the sea as described in Figure 4.

Step
8: the the
same
as in step 7,condition
replace thewhich
position
river
with the
sea which
achieves
the best
Step
9:
review
evaporation
canofbe
obtained
from
the pseudo
code.
solution. The river may provide a better solution than the sea as described in Figure 4.
Step 10: after the evaporation condition is attained, the precipitation process will start by using

Step 9: review the evaporation condition which can be obtained from the pseudo code.
Equations
(21)
and

Step 10:
after
the(22).
evaporation condition is attained, the precipitation process will start by using
Step 11:
decrease
by using Equation (20).
max (22).
Equations
(21)dand

Step
decrease
dmax bycriteria
using Equation
(20). the algorithm will be ended. Otherwise, return
Step
12: if11:
the
termination
are satisfied,

Step
12:
if
the
termination
criteria
are
satisfied,
the algorithm will be ended. Otherwise, return
back to step 5.
back to step 5.

The previous
procedures are summarized in the flowchart of the water cycle optimization
The previous procedures are summarized in the flowchart of the water cycle optimization
algorithm
illustrated
in Figure
5. 5.
algorithm
illustrated
in Figure
START

Introduce the system parameter,
and generate random initial
population (streams, rivers and
sea).

Determine the initial flow intensity utilizing the
Fitness Function (FF) calculations of each raindrop.

Streams flow to the rivers.
Rivers flow to the sea.

FFStream better than FFRiver

No

Yes

Exchange the stream position with the corresponding
river.

FFRiver better than FFSea

No

Yes

Exchange the river position with the sea.

Evaporation condition
is satisfied?

No

Yes
Create clouds, and start raining process,
then decrease the value of dmax

No

Convergence criteria
is fulfilled?

Yes
END

Figure 5. A flowchart for water cycle optimization algorithm.

Figure 5. A flowchart for water cycle optimization algorithm.

Energies 2018, 11, 1140

12 of 21

Energies 2018, 11, x FOR PEER REVIEW

12 of 22

GA is a well-known meta-heuristic technique. GA has been applied to different engineering
Energies
11, x FOR PEERmeta-heuristic
REVIEW
12 of 22
GAproblems.
is2018,
a well-known
technique.
GA has
beenon
applied
to different
engineering
technical
GA simulates the Darwin
theory
based
mating,
crossover,
and mutations.
technical
problems.
GA simulates
the Darwinthe
theory
on mating,becrossover,
and
mutations. Toto keep
To keep
theGA
best
generation,
elitebased
criteria
taken to
into
consideration
is chromosome
a well-known or
meta-heuristic
technique.
GA hasmust
been applied
different
engineering
keep the best chromosome or generation, the elite criteria must be taken into consideration to keep
the best
qualities.
The WCOA
results the
willDarwin
be compared
to GAonresults.
6 shows
the flowchart
technical
problems.
GA simulates
theory based
mating,Figure
crossover,
and mutations.
To of
the best qualities. The WCOA results will be compared to GA results. Figure 6 shows the flowchart
keep the
best chromosome
or generation, the elite criteria must be taken into consideration to keep
the Genatic
Algorithm
Technique.
of the Genatic Algorithm Technique.
the best qualities. The WCOA results will be compared to GA results. Figure 6 shows the flowchart
of the Genatic Algorithm Technique.
START

START
Initialize the first population
Initialize the first population
Calculate the fitness Function
Calculate the fitness Function
YES
Is the terminated criteria
satisfied

NO
NO

Output Display
YES

Is the terminated criteria
satisfied

Output Display

Selectiom
END
Selectiom
Crossover

END

Crossover
Mutation
Mutation

Figure
flowchart
of GA.
Figure6.6.A A
flowchart
of GA.
Figure 6. A flowchart of GA.
4. System Data under Study and Discussion

4. System Data under Study and Discussion

flowchart
minimizing
the Discussion
objective function using thermal (coal-fired) generating units,
4.A
System
Datafor
under
Study and
A flowchart
the objective
functioncost
using
generating
PEVs
and RERsfor
to minimizing
achieve reduction
in both operating
andthermal
support(coal-fired)
the decarbonization
in units,
A flowchart
for minimizing the
objective
function
using
(coal-fired)
generating units,
PEVs
andgrid
RERs
to achieve
both
operating
cost
andthermal
support
the decarbonization
in smart
smart
through
T = 24 reduction
h is shown in Figure
7.
PEVs and RERs to achieve reduction in both operating cost and support the decarbonization in
grid through T = 24 h is shown in Figure 7.
smart grid through T = 24 h is shown in Figure 7.

Figure 7. A flowchart for minimizing both cost and emission by using PEVs and RERs.
Figure 7. A flowchart for minimizing both cost and emission by using PEVs and RERs.

Figure 7. A flowchart for minimizing both cost and emission by using PEVs and RERs.

Energies 2018, 11, 1140

13 of 21

An independent system operator “ISO is designed of a single bus test system” which consists
of three thermal units designed for simulation with 5000 PEVs and RERs. Thermal generating units
are coal-fired due to their low operational costs; and the heat rates of (coal-fired) thermal units are
typically in the range of 9000 Btu/kWh to 11,000 with (fuel price = 1 $/MBtu). The thermal power
plant data is described and collected from [40] in Tables 2 and 3. Wind and solar power plants in this
model are described in Table 4 and data is collected from [41].
Table 2. Generator data.
Unit

Pmin (MW)

Pmax (MW)

Ramp up (MW/h)

Ramp down (MW/h)

Intial State at Time

G1
G2
G3

30
30
20

600
600
400

200
200
200

50
20
50

On
Off
On

Table 3. Generator energy data.
Fuel Consumption Function
Unit

A (MBtu)

B (MBtu/MWh)

C (MBtu/MWh2 )

Startup Cost
($)

Shut down Cost
($)

176.9
129.9
137.4

13.5
40.6
17.6

0.04
0.001
0.005

1200
1000
1500

800
500
800

G1
G2
G3

Table 4. Wind–solar energy.
Hour

Wind
(MW)

Solar
(MW)

Hour

Wind
(MW)

Solar
(MW)

Hour

Wind
(MW)

Solar
(MW)

Hour

Wind
(MW)

Solar
(MW)

1
2
3
4
5
6

8.2
11.4
66.9
69.8
55.4
50.9

0
0
0
0
0
0

7
8
9
10
11
12

4.6
49.3
45.6
10.1
24.8
37.3

5
22.04
53.95
67.4
67.32
69.64

13
14
15
16
17
18

59.0
78.1
44.9
19.5
3.7
16.5

65.0
58.27
53.79
47.06
27.11
11

19
20
21
22
23
24

72.2
73.3
65.3
24.5
49.9
40.3

0
0
0
0
0
0

The parameters of a random model of PEVs are: Expected total number of PEVs in the smart
grid = 5000; Maximum capacity of the battery = 25 kWh; Minimum capacity of the battery = 10 kWh;
Average capacity of the battery “Pavg” = 15 kWh; Frequency of charging/discharging =1 per day;
Departure state of charge (SOC) “Ψdep ” = 50%; Efficiency “η” = 85%. A typical PEV needs about
8.22 kWh/day (41.1 MWh/day) for 5000 vehicles, 2.6 $/gallon (fuel price = 1 $/MBtu) for gasoline
price taking into consideration the emission factor for fuel oil from Table 1, assuming the scenario for
simulation of PEVs according to the demand which is relatively low during hours 1–7, 16–19 and 23–24
(a total of 12 h). PEVs can be charged during the off-peak load. Therefore, an additional 41.1 MWh
load for 5000 vehicles assuming the number of vehicles will be charged for each hour in Figure 8 [41].
Costs for wind and solar energy are estimated by the International Renewable Energy Agency
(IRENA). The levelized cost of electricity (LCOE) from solar photovoltaics (PV) fell by 69% between
2010 and 2017 reaching the cost range of fossil fuels. Wind costs fell by 15% in the same period.
For solar energy (14.597 $/MW) and wind energy (10 $/MW) [42].

Energies 2018, 11, 1140

14 of 21

Energies 2018, 11, x FOR PEER REVIEW

14 of 22

Figure 8. Load leveling for PEVs. Note: The PEVs is integrated as load for the whole day except from
hourleveling
8 till hourfor
15 and
from Note:
hour 20 till
hour
22. is integrated as load for the whole day except from
Figure 8. Load
PEVs.
The
PEVs
hour 8 till hour 15 and from hour 20 till hour 22.

5. Simulation Results

The parameters for the proposed algorithm (WCOA): Npop = 70, Nsr = 3 (2 rivers + 1 sea), dmax

5. Simulation
Results
−5, max_iteration = 700, assuming the spinning reserve is a minimum of 10% of the load demand
= 10

condition. Four cases are being considered and the results from the proposed algorithm (WCOA) are

The parameters
for the proposed algorithm (WCOA): N (DP).=Tables
70, N
= 3 explain
(2 rivers + 1 sea),
sr below
compared with Genetic Algorithm (GA) and Dynamic Programmingpop
5–8
dmax = 10−5applying
, max_iteration
= 700,
assuming
spinning
reserve
is aDP,
minimum
of 10%
the WCOA. The
whole
comparisonthe
of the
three applied
techniques
GA, and WCOA
is of the load
shown
in
Table
9.
demand condition. Four cases are being considered and the results from the proposed algorithm
(WCOA) are5.1.
compared
Case 1 (Base with
Case) Genetic Algorithm (GA) and Dynamic Programming (DP). Tables 5–8
below explain applying
the WCOA. The whole comparison of the three applied techniques DP, GA,
Case 1 illustrates the distribution of the unit commitment scheduling for the three thermal units
and WCOA is
shown
ininTable
9.
only
as shown
Table 5.
5.1. Case 1 (Base Case)
Time

Table 5. Base case of 3 thermal Generating units’ applying the WCOA.
ThermUnit-1

ThermUnit-2

ThermUnit-3

Emission

Demand

Case 1 illustrates
of the
unit commitment
scheduling
for(MW)
the three thermal units
(Hour)the distribution
(MW)
(MW)
(MW)
(ton)
1
67.735
0
132.265
191
200
only as shown in Table
5.
2
67.778
0
132.222
191
200
3
73.334
0
176.666
238.75
250
Table
5. Base
case of 3 thermal
units’ applying
4
73.333
0 Generating
176.667
238.75 the WCOA.
250
5
73.332
0
176.668
238.75
250
6
67.633
0
132.367
191
200
Time
ThermUnit-3
Demand
ThermUnit-1
(MW)
ThermUnit-2
(MW)
Emission (ton)350
84.444
0
265.556
334.25
(Hour) 7
(MW)
(MW)
8
101.109
0
398.891
477.5
500
1
67.735
0
132.265
191
200
9
92.734
158.375
348.891
573
600
2
67.778
0
132.222
191
200
10
97.089
329.012
373.899
764
800
3
73.334
0
176.666
238.75
250
11
96.688
339.075
364.237
764
800
4
73.333
0
176.667
238.75
250
12
66.688
319.075
314.237
668.5 238.75
700
5
73.332
0
176.668
250
6
67.633
0
132.367
200
13
95.658
299.075
355.267
716.25 191
750
7
84.444
0
265.556
350
14
95.126
296.051
358.823
716.25 334.25
750
8
101.109
0
398.891
500
15
92.654
276.051
331.295
668.5 477.5
700
9
92.734
158.375
348.891
600
16
89.327
256.051
304.622
620.75 573
650
10
97.089
329.012
373.899
764
800
17
101.11
0
398.89
477.5
500
11
96.688
339.075
364.237
764
800
18
92.677
158.433
348.89
573
600
12
66.688
319.075
314.237
668.5
700
13
95.658
299.075
355.267
716.25
750
14
95.126
296.051
358.823
716.25
750
15
92.654
276.051
331.295
668.5
700
16
89.327
256.051
304.622
620.75
650
17
101.11
0
398.89
477.5
500
18
92.677
158.433
348.89
573
600
19
93.117
173.822
333.061
573
600
20
95.146
254.079
350.775
668.5
700
21
91.782
234.08
324.138
620.75
650
22
150
0
400
525.25
550
23
100
0
350
429.75
450
24
50
0
300
334.25
350
Start-up cost

Fuel cost

Total cost

Emissions

3000.000 $/day

247,284.867 $/day

368,227.367 $/day

11,794.250 ton/day

The total production and emission cost for the base three generator units is 368,227.367 $/day.

Energies 2018, 11, 1140

15 of 21

5.2. Case 2 (PEVs and Three Thermal Units)
PEVs represent one of the newly added techniques to gain energy to support the network.
Many strategies were investigated, and most of them agreed on applying PEVs to reduce small
fluctuations and improve quality [41]. Case 2 represents the unit commitment after integrating 5000
PEV (V2G/G2V) with the three thermal units, as shown in Table 6.
Table 6. Integration PEVs (V2G/G2V) with base 3 thermal generating units’ applying the WCOA.
Time
(Hour)

ThermUnit-1
(MW)

ThermUnit-2
(MW)

ThermUnit-3
(MW)

PEV Unit-4
(V2G/G2V)

Emission (Ton)

Demand
(MW)

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24

68.173
68.174
73.677
73.659
73.531
68.852
84.852
100.221
192.733
339.856
339.777
289.777
311.075
306.945
264.993
222.242
0
120.136
202.808
291.596
241.596
191.596
141.596
0

0
0
0
0
0
0
0
0
0
44.207
41.098
30
30
30
30
30
102.363
82.363
0
0
0
0
0
0

135.381
135.342
179.391
179.554
179.481
135.454
268.82
391.791
400
400
400
364.285
400
400
400
400
400
400
400
400
396.849
351.682
312.563
353.322

−3.554
−3.516
−3.068
−3.213
−3.012
−3.624
−3.672
7.988
7.268
15.938
19.125
15.938
8.925
13.055
5.007
−2.242
−2.363
−2.499
−2.808
8.404
11.555
6.721
−4.159
−3.322

194.394
194.358
241.68
241.818
241.626
194.461
337.757
477.005
572.549
763.012
762.814
667.512
715.697
715.441
668.19
622.891
479.756
575.386
575.681
667.979
620.034
524.833
433.722
337.422

203.554
203.516
253.068
253.213
253.012
203.624
353.672
500
600
800
800
700
750
750
700
652.242
502.363
602.499
602.808
700
650
550
454.159
353.322

Start-up cost

Fuel cost

Total cost

Emissions

4300.000 $/day

269,843.179 $/day

392,403.355 $/day

11,826.018 ton/day

Note: load is leveled when the load of PEVs (G2V) is added to demand, Positive and negative values indicate
V2G/G2V (discharging/charging) respectively.

By comparing both Tables 6 and 7, start-up, production cost, and emission increase when
integrating PEVs with the base case three thermal units by 1300 $/day; 22,558.31 $/day and exceeding
the emission by 31.768 ton/day from the coal-fired generators to supply energy to PEVs and emissions
from PEVS 90.7441 ton/day (33,121.597 ton/year for 5000 vehicles) in the transportation sector [34].
5.3. Case3 (RERs, and Three Thermal Units)
Integrating RERs, both wind and solar energy to partially replace thermal (coal-fired) generation
units will contribute to reduction in both the production and the emission cost. However, production
costs of wind and solar energy resources are relatively high; and the operation and maintenance prices
are significantly decreasing. This encourages the usage of these energy resources that also support
decarbonizing in the smart grid. Case 3 signifies the unit commitment after integrating RERs (wind &
solar), with the three thermal units, as shown in Table 7.

Energies 2018, 11, 1140

16 of 21

Table 7. Integration of RESs (Wind-Solar) with base three thermal generating units’ simulation results
by using the WCOA.
Time
(Hour)

ThermUnit-1
(MW)

ThermUnit-2
(MW)

ThermUnit-3
(MW)

Wind
(MW)

Solar
(MW)

Emission
(ton)

Demand
(MW)

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24

66.867
66.511
65.901
65.578
67.178
62.122
83.378
93.132
101.16
292.5
277.88
0
0
0
57.68
183.44
133.44
172.8
127.8
226.7
184.7
134.7
95.556
45.556

0
0
0
0
0
0
0
0
0
30
30
193.06
226
213.63
193.63
0
0
0
0
0
0
0
0
0

124.933
122.089
117.119
114.622
127.422
86.978
257.022
335.528
399.29
400
400
400
400
400
350
400
362.86
400
400
400
400
390.8
354.444
304.444

8.2
11.4
66.9
69.8
55.4
50.9
4.6
49.3
45.6
10.1
24.8
37.3
59
78.1
44.9
19.5
3.7
16.5
72.2
73.3
65.3
24.5
0
0

0
0
0
0
0
0
5
22.04
53.95
67.4
67.32
69.64
65
58.27
53.79
47.06
0
11
0
0
0
0
0
0

183.341
180.352
176.265
173.557
187.006
143.459
325.209
410.538
479.211
690.604
676.95
567.573
599.459
588.006
575.517
557.877
474.044
547.15
505.565
600.038
559.76
502.367
429.75
334.25

200
200
250
250
250
200
350
500
600
800
800
700
750
750
700
650
500
600
600
700
650
550
450
350

Start-up cost

Fuel cost

Total cost

Emissions

4800.000 $/day

257,120.030 $/day

366,598.528 $/day

10,467.850 ton/day

Integrating both RERs and PEVs into the smart grid showed a significant decrease in fuel
cost, emissions and total cost which fell to 5,315.365 $/day, 40.567 ton/day; 3,420.759 $/day as
compared to integrating RESs with the three thermal units which led to decrease to 18,038.244 $/day,
1,398.735 ton/day, 29,225.586 $/day and compared to integrating PEVs into the three thermal units.
5.4. Case 4 (RERs, PEVs, and the Three Thermal Units)
RERs are accumulated to reduce emissions from both conventional units and PEVs. On the other
hand, PEVs are integrated to solve the uncertainty behavior of RERs. Therefore, further enhancement
for the reliability and stability of the grid can be covered up against any unexpected uncertainty
behavior for RESs by including the grid for both the integration of RES and PEVs. Case 4 implies the
unit commitment after integrating RERs (wind & solar), 5000 PEV (V2G/G2V) with the 3 thermal
units, as shown in Table 8.
Comparing the results obtained from the proposed algorithm (WCOA) with the results obtained
from the forward dynamic programming (DP) and genetic algorithm (GA) as shown in Table 9
including the emission and emission cost.
Table 9 sums up the whole comparison between the results obtained from the proposed algorithm
(WCOA) with the results obtained from (DP) and (GA). Table 9 provides three main outcomes. First,
the power of the WCOA in solving the unit commitment problem indicated in the yellow boxes.
Second, the green box shows that the emission cost slightly increased integrating combined PEVs but
with the existence of PEVs to support wind and solar units. So the need to pay more attention and
facilities to invest in the PEVs (electrically based) will lead to more reduction in the emission costs.
Third, focusing on the production cost in any case aims to lower the price to 2/3 of its price, but this
will increase cost on the long term operation.

Energies 2018, 11, 1140

17 of 21

Table 8. Integration PEVs with RESs (wind-solar) with base three thermal generating units’ simulation
results by the WCOA.
Time
(Hour)

Therm
Unit-1
(MW)

Therm Unit-2
(MW)

Therm
Unit-3
(MW)

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24

67.262
66.902
66.283
65.855
67.511
62.524
83.785
92.344
100.351
276.563
258.755
0
96.516
97.064
196.303
185.682
0
123.446
101.006
188.296
143.145
99.866
65.346
64.409

0
0
0
0
0
0
0
0
0
30
30
177.123
157.123
137.123
0
0
71.553
51.553
31.553
30
30
30
0
0

128.092
125.214
119.885
117.558
130.1
90.2
260.287
328.328
392.831
400
400
400
363.437
366.389
400
400
400
400
398.049
400
400
388.913
338.913
288.913

PEV Unit-4
(V2G/G2V)(MW)

Wind
(MW)

Solar
(MW)

Emission
(Ton)

Demand
(MW)

8.2
11.4
66.9
69.8
55.4
50.9
4.6
49.3
45.6
10.1
24.8
37.3
59
78.1
44.9
19.5
3.7
16.5
72.2
73.3
65.3
24.5
49.9
0

0
0
0
0
0
0
5
22.04
53.95
67.4
67.32
69.64
65
58.27
53.79
47.06
27.11
11
0
0
0
0
0
0

186.735
183.71
179.195
176.625
189.882
146.92
328.715
410.043
478.76
689.616
675.764
566.585
598.906
587.197
575.206
560.018
450.573
549.536
508.246
599.517
559.043
501.95
387.115
337.422

203.554
203.516
253.068
253.213
253.012
203.624
353.672
500
600
800
800
700
750
750
700
652.242
502.363
602.499
602.808
700
650
550
454.159
353.322

−3.554
−3.516
−3.068
−3.213
−3.012
−3.624
−3.672
7.988
7.268
15.938
19.125
15.938
8.925
13.055
5.007
−2.242
−2.363
−2.499
−2.808
8.404
11.555
6.721
−4.159
−3.322

Start-up cost

Fuel cost

Emissions

Total cost

7100.000 $/day

251,804.935 $/day

10,427.283 ton/day

363,177.769 $/day

Note: load is leveled when the load of PEVs (G2V) is added to demand, Positive and negative values indicate
V2G/G2V (discharging/charging) respectively.

Table 9. Results summary for WCOA, GA and DP (including emission cost).

Algorithm

WCOA

GA

DP

Modes

Start-Up
Cost
($/Day)

Production
Cost
($/Day)

Total Cost
(Including Emission
Cost) ($/Day)

Emissions
(Ton/Day)

Base 3 thermal units

3000

247,284.867

368,227.367

11,794.250

PEVs(V2G/G2V) with
3 thermal units

4300

269,843.179

392,403.355

11,826.018

RESs with 3-thermal units

4800

257,120.30

366,598.528

10,467.850

PEVs(V2G/G2V), RESs
with 3-thermal units

7100

251,804.935

363,177.769

10,427.283

Base 3 thermal units

3000

247,284.949

368,227.222

11,794.227

PEVs(V2G/G2V) with
3 thermal units

4300

272,676.166

395,270.780

11,829.461

RESs with 3-thermal units

4800

264,149.704

376,118.886

10,716.918

PEVs(V2G/G2V), RESs
with 3 thermal units

4400

258,840.379

367,602.876

10,436.250

Base 3-thermal units

3000

247,281.115

368,223.615

11,794.250

PEVs(V2G/G2V) with
3 thermal units

4300

270,155.998

392,716.173

11,826.018

RESs with 3-thermal units

4800

257,120.030

366,598.528

10,467.850

PEVs(V2G/G2V), RESs
with 3 thermal units

7100

251,804.939

363,177.773

10,427.283

Energies 2018, 11, 1140

18 of 21

6. Discussion
The above mentioned tables can be summed up as follows:








The total cost and emissions for the base 3-generator units increase when integrating PEVs with
the base 3 thermal units, and exceed the emission limit more than the base case.
Integrating RERs, both wind and solar energy to partially replace thermal (coal-fired) generation
units contributes to reducting both the fuel cost and the emissions. Significant reduction in
emission occurs when integrating only RERs with the base three thermal units. However,
the uncertainty of wind and solar energy is based on several factors such as geographical area,
the forecasting models used and period-ahead forecasting. These variables affect the uncertainty
percentage and the overall accuracy.
Integrating both RERs and PEVs into the smart grid showed a significant decrease in fuel cost,
emissions and total cost. RERs is accumulated to reduce emissions from both conventional units
and PEVs. PEVs is integrated to solve the uncertainty behavior of RERs.
The results extracted from the proposed algorithm (WCOA) proved its efficiency with respect to
the results of both dynamic programing (DP) and genetic algorithm (GA).

7. Conclusions
In this paper, the unit commitment problem has been solved in four different modes with three
algorithms, provided with results and analysis to achieve;
(i) Reduction in total cost including the emission cost.
(ii) Decarbonizing the emissions from the conventional generators and the transportation sector.
(iii) Presenting new types of storage energy in the electricity sector by replacing conventional vehicles
with electrical vehicles with environment-friendly batteries and encouraging the consumers to
supply electrical power to the grid during the on-peak periods.
The four modes are;
(1)
(2)
(3)
(4)

The base 3-thermal (coal-fired) units.
PEVs (V2G/G2V) with the conventional units.
RERs (wind/Solar) with the conventional units.
Integrating PEVs (V2G/G2V), RERs (wind/solar) with the conventional units.
The four modes are executed applying three algorithms which are;

(1)
(2)
(3)

Water cycle optimization algorithm (WCOA).
Genetic algorithm (GA).
Dynamic programing (DP).

Based on the comparison of the proposed technique with other optimization algorithms, WCOA
has shown promising performance and better solutions than GA and DP techniques. In the four
modes of study, the WCOA offers competitive results with respect to other meta-heuristic optimization
techniques with acceptable degree of accuracy for the solutions.
From this study, it is concluded that, integrating both PEVs and RERs with the conventional
generating units achieved many purposes;
(i) Increasing the reliability and stability of the electricity grid.
(ii) Decarbonizing the emissions from the electricity sector and transportation sector.
(iii) Introducing new types of unit commitment sourses with different distributions, for environmentfriendly electrical energy storage such as PEVs (V2G) to encourage the consumers to supply
electrical power to the grid during the on-peak periods of the electrical network operation.

Energies 2018, 11, 1140

19 of 21

Author Contributions: The four authors contribute in the whole work. H.-A.I.E., R.A.S. and N.H.E.-A. worked
on the system modelling and the optimization techniques analysis and implementation. H.K.T. supervised the
paper writing and reviewing.
Conflicts of Interest: The authors declare no conflicts of interest.

Nomenclature
PG i
A, B, C
NG
Pwind
Psolar
α
β
PPEVj
η
NV2G
NV2G max
N
ΨPres
Ψdep
Ψmin
Ψmax
Ui (hour)
X
N
2N − 1
Nvar
Raindrop
(X1 , X2 , X3 , . . . , XNar )
NSn
Npop
Nsr
X
LB and UB
dmax
µ

the output power of each thermal unit “i” at each hour
the coefficients of a quadratic fuel cost function of each thermal generating unit
the number of conventional thermal units
the output power from a wind plant at each hour
the output power from a solar plant at each hour
CO2 emission factor
the emission penalty factor
the power of each vehicle j
the system efficiency
number of vehicles that are connected to the network at this hour
the total vehicles in the network
number of units that are on in the unit commitment problem at each hour
the present state of charge (SOC)
the departure state of charge (SOC)
the depletion of storage energy at minimum level
the charging up to maximum level
on/off state of each unit “i”
the number of states to search each period in DP algorithm
the number of strategies to save at each step in DP algorithm
maximum value of X or N in DP algorithm
a dimensional optimization problem or number of design variables in the WCO algorithm
a single solution in an array of 1 × Nvar in the WCO algorithm
the decision variable values in WCO algorithm
the number of streams that travel towards certain rivers or the sea in the WCO algorithm
the number of population in the WCO algorithm
the summation of the number of rivers in the WCO algorithm
the distance between the stream and the river in the WCO algorithm
lower and upper boundaries of the designed problem in the WCO algorithm
a small number and its value is near to zero in the WCO algorithm
a coefficient to get the range of searching area close to the sea in the WCO algorithm

References
1.
2.

3.

4.

5.

Lopez, C.J.; Ano, O.; Esteybar, D.O. Stochastic Unit Commitment & Optimal Allocation of Reserves: A Hybrid
Decomposition Approach. IEEE Trans. Power Syst. 2018. [CrossRef]
Krishnamurthy, S.; Tzoneva, R. Comparative analyses of Min-Max and Max-Max price penalty factor
approaches for multi criteria power system dispatch problem with valve point effect loading using
Lagrange’s method. In Proceedings of the 2011 IEEE International Conference on Power and Energy
Systems (ICPS), Chennai, India, 22–24 December 2011. [CrossRef]
Zhou, B.; Ai, X.; Fang, J.; Wen, J.; Yang, J. Mixed-integer second-order cone programming taking appropriate
approximation for the unit commitment in hybrid AC–DC grid. In Proceedings of the IEEE 6th International
Conference on Renewable Power Generation (RPG), Wuhan, China, 19–20 October 2017.
Alejandre, L.M.C.; Alcaraz, G.G. Analysis of Security Constrained Unit Commitment using Three Models
of Electricity Generation Cost Linearization. In Proceedings of the IEEE Conference on Texas Power and
Energy Conference (TPEC), College Station, TX, USA, 8–9 February 2018. [CrossRef]
Elsayed, A.M.; Maklad, A.M.; Farrag, S.M. A New Priority List Unit Commitment Method for Large-Scale
Power Systems. In Proceedings of the 2017 Nineteenth International Middle East Power Systems Conference
(MEPCON), Cairo, Egypt, 19–21 December 2017.

Energies 2018, 11, 1140

6.

7.
8.
9.
10.

11.
12.

13.
14.

15.
16.

17.
18.
19.
20.
21.
22.

23.
24.
25.
26.

27.

20 of 21

Venayagamoorthy, G.K.; Braband, G. Carbon reduction potential with intelligent control of power systems.
In Proceedings of the 17th World Congress: International Federation of Automatic Control, Seoul, Korea,
6–11 July 2008.
Labatt, S.; White, R.R. Carbon Finance: The Financial Implications of Climate Change; Wiley: Hoboken, NJ,
USA, 2007.
Godina, R.; Rodrigues, E.M.G.; Pouresmaeil, E.; Catalão, J.P.S. Optimal residential model predictive control
energy management performance with PV microgeneration. Comput. Oper. Res. 2017, in press.
Rodrigues, E.M.G.; Godina, R.; Pouresmaeil, E.; Ferreira, J.R.; Catalão, J.P.S. Domestic appliances energy
optimization with model predictive control. Energy Convers. Manag. 2017, 142, 402–413. [CrossRef]
Oliveira, D.; Rodrigues, E.M.G.; Godina, R.; Mendes, T.D.P.; Catalão, J.P.S.; Pouresmaeil, E. Enhancing Home
Appliances Energy Optimization with Solar Power Integration. In Proceedings of the IEEE International
Conference on Computer as a Tool (EUROCON 2015), Salamanca, Spain, 8–11 September 2015. [CrossRef]
Godina, R.; Rodrigues, E.M.G.; Pouresmaei, E.; Matias, J.C.O.; Catalão, J.P.S. Model Predictive Control Home
Energy Management and Optimization Strategy with Demand Response. Appl. Sci. 2018, 8, 408. [CrossRef]
Silva, J.M.; Rodrigues, E.; Godina, R.; Catalao, J.P.S. Residential MPC controller performance in a household
with PV microgeneration. In Proceedings of the 2017 IEEE Conference on Manchester PowerTech, Manchester,
UK, 18–22 June 2017. [CrossRef]
Nguyen, D.T.; Le, L.B. Risk-constrained profit maximization for microgrid aggregators with demand
response. IEEE Trans. Smart Grid 2015, 6, 135–146. [CrossRef]
Stock, A.; Stock, P.; Sahajwalla, V. Powerful Potential: Battery Storage for Renewable Energy and Electric
Cars; Climate Council of Australia Ltd.: Sydney, Australia, 2015; ISBN 978-0-9944195-4-5 (print),
978-0-9944195-3-8 (web).
Nieto, A.; Vita, V.; Maris, T.I. Power quality improvement in power grids with the integration of energy
storage systems. Int. J. Eng. Res. Technol. (IJERT) 2016, 5, 438–443.
Nykvist, B.; Nilsson, M. Rapidly falling costs of battery packs for electric vehicles. Nat. Clim. Chang. 2015,
5, 329–332. Available online: http://www.nature.com/nclimate/journal/v5/n4/full/nclimate2564.html
(accessed on 12 March 2018). [CrossRef]
King Island Advanced Hybrid Power Station. Available online: www.kingislandrenewableenergy.com.au/
history/king-island-renewable-energy (accessed on 13 March 2018).
Al Ghaithi, H.M.; Fotis, G.P.; Vita, V. Techno-economic assessment of hybrid energy off-grid system—A case
study for Masirah island in Oman. Int. J. Power Energy Res. 2017, 1, 103–116. [CrossRef]
Agamah, S.U.; Ekonomou, L. Energy storage system scheduling for peak demand reduction using
evolutionary combinatorial optimization. Sustain. Energy Technol. Assess. 2017, 23, 73–82.
Agamah, S.; Ekonomou, L. A heuristic combinatorial optimization algorithm for load-leveling and peak
demand reduction using energy storage systems. Electr. Power Compon. Syst. 2018, 45, 2093–2103. [CrossRef]
Center for Climate and Energy Solution. Climate Solution, Technology Solution, Electrical Vehicles. Available
online: www.c2es.org/content/electric-vehicles/ (accessed on 3 March 2018).
Little, A.D.; Browning, L.; Santini, D.; Vyas, A.; Taylor, D.; Markel, T.; Duvall, M.; Graham, R.; Miller, A.;
Frank, A.; et al. Comparing the Benefits and Impacts of Hybrid Electric Vehicle Options; for Compact Sedan and Sport
Utility Vehicles; Electric Power Research Institute (EPRI): Palo Alto, CA, USA, 2002; p. 1006892. Available
online: http://www.evworld.com/library/EPRI_sedan_options.pdf (accessed on 13 March 2018).
Corrigan, D.; Masias, A. Batteries for Electric and Hybrid Vehicles. In Linden’s Handbook of Batteries, 4th ed.;
Reddy, T.B., Ed.; McGraw Hill: New York, NY, USA, 2011.
Vita, V. Development of a decision-making algorithm for the optimum size and placement of distributed
generation units in distribution networks. Energies 2017, 10, 1433. [CrossRef]
Boulanger, A.G.; Chu, A.C.; Maxx, S.; Waltz, D.L. Vehicle electrification: Status and issues. Proc. IEEE 2011,
99, 1116–1138. [CrossRef]
Ghasemi, A.; Mortazavi, S.S.; Mashhour, E. Hourly demand response and battery energy storage for
imbalance reduction of smart distribution company embedded with electric vehicles and wind farms.
Renew. Energy 2016, 85, 124–136. [CrossRef]
Morais, H.; Sousa, T.; Soares, J.; Faria, P.; Vale, Z. Distributed energy resources management using plug-in
hybrid electric vehicles as a fuel-shifting demand response resource. Energy Convers. Manag. 2015, 97, 78–93.
[CrossRef]

Energies 2018, 11, 1140

28.

29.
30.
31.

32.
33.
34.
35.
36.
37.

38.
39.

40.

41.
42.

21 of 21

Li, K.; Xue, Y.; Cui, S.; Niu, Q.; Yang, Z.; Luk, P. Advanced computational methods in energy, power, electric
vehicles and their integration. In Proceedings of the International Conference on Life System Modeling and
Simulation (LSMS2017) and International Conference on Intelligent Computing for Sustainable Energy and
Environment (ICSEE 2017), Nanjing, China, 22–24 September 2017; Part III. Springer: Berlin, Germany, 1842.
Wedyan, A.; Whalley, J.; Narayanan, A. Hydrological Cycle Algorithm for Continuous Optimization
Problems. J. Glob. Optim. 2017, 37, 405–436. [CrossRef]
Moradi, M.; Sadollah, A.; Eskandar, H.; Eskandarc, H. The application of water cycle algorithm to portfolio
selection. J. Econ. Res. Ekon. Istraž. 2017, 30, 1277–1299.
Yanjun, K.; Yadong, M.; Weinan, L.; Xianxun, W.; Yue, B. An enhanced water cycle algorithm for optimization
of multi-reservoir systems. In Proceedings of the 2017 IEEE/ACIS 16th International Conference on
Computer and Information Science (ICIS), Wuhan, China, 24–26 May 2017.
Eskandara, H.; Sadollahb, H.; Bahreininejadb, A. Weight Optimization of Truss Structures Using Water Cycle
Algorithm. Int. J. Optim. Civ. Eng. 2013, 3, 115–129.
Sadollah, A.; Eskandar, H.; Kim, J. Water cycle algorithm for solving constrained multi-objective optimization
problems. Appl. Soft Comput. 2015, 27, 279–298. [CrossRef]
World Bank and Ecofys. Carbon Pricing Watch 2017. 2017. Available online: https://openknowledge.
worldbank.org/handle/10986/26565, (accessed on 3 March 2018).
Tung, N.S.; Bhadoria, A.; Kaur, K.; Bhadauria, S. Dynamic programming model based on cost minimization
algorithms for thermal generating units. Int. J. Enhanc. Res. Sci. Technol. Eng. 2012, 1, 19–27.
Thakur, N.; Titare, L.S. Determination of unit commitment problem using dynamic programming. Int. J. Nov.
Res. Electr. Mech. Eng. 2016, 3, 24–28.
Shi, J.; Dong, X.; Zhao, T.; Du, Y.; Liu, H.; Wang, Z.; Zhu, D.; Xiong, C.; Jiang, L.; Shi, J.; et al. The Water Cycle
Observation Mission (WCOM): Overview. In Proceedings of the IEEE International Geoscience and Remote
Sensing Symposium (IGARSS), Beijing, China, 10–15 July 2016.
Bozorg-Haddad, O.; Solgi, M.; Loáiciga, H.A. Meta-Heuristic and Evolutionary Algorithms for Engineering
Optimization, 1st ed.; John Wiley & Sons, Inc.: Hoboken, NJ, USA, 2017; pp. 231–240.
Eskandar, H.; Sadollah, A.; Bahreininejad, A.; Hamdi, M. Water cycle algorithm—A novel metaheuristic
optimization method for solving constrained engineering optimization problems. Comput. Struct. 2012,
110–111, 151–166. [CrossRef]
He, D.; Tan, Z.; Harley, R.G. Chance constrained unit commitment with wind generation and
superconducting magnetic energy storages. In Proceedings of the IEEE Power and Energy Society General
Meeting, San Diego, CA, USA, 22–26 July 2012.
Saber, A.Y.; Venayagamoorthy, G.K. Efficient utilization of renewable energy sources by gridable vehicles in
cyber-physical energy systems. IEEE Syst. J. 2010, 4, 285–294. [CrossRef]
International Renewable Energy Agency (IRENA). Renewable Power Generation Costs in 2017; International
Renewable Energy Agency: Abu Dhabi, United Arab Emirates, 2018; Available online: http://www.irena.
org/publications/2018/Jan/Renewable-power-generation-costs-in-2017 (accessed on 18 March 2018).
© 2018 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).


Related documents


PDF Document untitled pdf document 4
PDF Document reducing emissions and improving energy efficiency
PDF Document 52i18 ijaet0514309 v6 iss6 2766 2783
PDF Document applications of energy storage technology
PDF Document j env hydrol 12 2004 1 11
PDF Document cheap solar panels   where to get them


Related keywords